Sentiment analysis using Support Vector Machines (SVM) on textual data. Sentiment analysis is a natural language processing (NLP) technique used to determine the sentiment (positive, negative, or neutral) expressed in a piece of text.
The script utilizes the popular scikit-learn library for machine learning and NLP tasks. It demonstrates how to preprocess textual data, vectorize it using Term Frequency-Inverse Document Frequency (TF-IDF) representation, train a Linear Support Vector Machine classifier, and make predictions on new user input.
Ensure you have the following Python libraries installed:
- scikit-learn
- numpy
- pandas
- nltk
You can install the required dependencies using pip:
pip install scikit-learn numpy pandas nltkMake sure you have a CSV file named Sentimental Analysis Data.csv
with two columns: text and sentiment. The text column should
contain the textual data, and the sentiment column should contain the
corresponding sentiment labels (positive, negative, neutral).
The text data undergoes thorough preprocessing before training and prediction. The preprocessing steps include:
- Removing punctuation, numbers and special characters.
- Converting text to lowercase.
- Tokenizing the text.
- Applying stemming using Porter Stemmer.
These steps help to ensure that the text data is in a suitable format for the
SVMclassifier.
The script uses the TfidfVectorizer to convert the text data into a numerical
representation based on the TF-IDF algorithm. Subsequently, a Linear Support
Vector Machine (LinearSVC) classifier is trained on the vectorized data to
predict the sentiment labels.
To assess the model's performance, the script evaluates its predictions using precision, recall, accuracy, and F1 measure. The evaluation is conducted by predicting the sentiment label for user input and comparing it with the actual label.
This project is licensed under the MIT License - see the LICENSE file for details.